Acta Psychiatr Scand 2002: 105: 346±355 Printed in UK. All rights reserved
Copyright Ó Blackwell Munksgaard 2002 ACTA PSYCHIATRICA SCANDINAVICA ISSN 0001-690X
Comparisons between schizophrenia patients recruited from Australian general practices and public mental health services Carr VJ, Lewin TJ, Barnard RE, Walton JM, Allen JL, Constable PM, Chapman JL. Comparisons between schizophrenia patients recruited from Australian general practices and public mental health services. Acta Psychiatr Scand 2002: 105: 346±355. Ó Blackwell Munksgaard 2002. Objective: To examine dierences between samples of schizophrenia patients recruited from general practice and public mental health services. Method: Demographic, psychosocial, disability and 12-month service utilization data are reported from a multicentered survey of psychotic disorders and an associated study of schizophrenia in general practice. Patients with schizophrenia from three recruitment sources (in-patient, community services, general practice) were compared. Results: General practice patients had fewer symptoms, better functioning, lower service use, but comparable substance abuse, to patients from mental health services. They were generally similar to community mental health patients, with the exception of family support, premorbid work adjustment, negative symptoms and disability. Service contact models are also reported which demonstrate that general practitioners deal with schizophrenia patients across the range of illness severity and acuity. Conclusion: Recruitment source impacts in schizophrenia research need to be more carefully considered during sample selection and better accounted for in the interpretation of results.
Introduction
As schizophrenia has increasingly been treated in community settings the role of the general practitioner (GP) has assumed greater importance for this group of patients. Several innovative models of care for people with schizophrenia using GPs have been developed (1±5). In Australia, three-quarters of GPs care for people with schizophrenia, the typical GP treating three such patients, two conjointly with specialist services and one without this support (6). Thus, GPs have assumed a collective responsibility for the treatment of schizophrenia that is suciently large to warrant the development of means to evaluate and enhance the treatment of schizophrenia in this setting, in contrast to traditional mental health services. Relatively little is known, however, about the characteristics of schizophrenia patients treated primarily in general practice compared with those 346
Vaughan J. Carr1, Terry J. Lewin2, Rosemary E. Barnard3, Jane M. Walton3, Jennifer L. Allen3, Paul M. Constable3, Jenny L. Chapman3 1
Discipline of Psychiatry, University of Newcastle Hunter Mental Health, and 3Centre for Mental Health Studies, University of Newcastle, Newcastle, Callaghan, Australia
2
Key words: Australia; family practice; mental health services; patient selection; research design; schizophrenia Vaughan J. Carr, Centre for Mental Health Studies, University of Newcastle, Callaghan, NSW 2308, Australia E-mail:
[email protected] Accepted for publication 23 October, 2001
treated primarily in public mental health services. Potential dierences in socio-demographic pro®le, symptoms, disability, social support, patterns of comorbidity, service utilization, and unmet need across treatment settings, have yet to be fully investigated. A closely related issue, in the study of schizophrenia generally, is the question of how representative of the population of people with schizophrenia are samples recruited from public mental health services (in-patient and out-patient), general practices and other sources. Research ®ndings based on one sample source may not be generalizable to other sample sources or the entire population of people with schizophrenia, particularly if sample sizes are small and various uncontrolled factors (e.g. motivation, capacity to consent, etc.) in¯uence participant selection. No studies reported in the literature have directly compared schizophrenia patients recruited from general practice and from mental health
Schizophrenia and recruitment source services using the same inclusion and measurement criteria. Hence, any comparisons based on these sources have to consider similar but not matching measures. Also, as most studies reporting characteristics of patients recruited through GPs were conducted in the UK, where primary care and mental health services tend to be closely integrated (7), extrapolating ®ndings to other health care systems must be undertaken with caution. Schizophrenia samples recruited through mental health services appear to be younger, and to have higher proportions of males and single persons than those recruited through GPs (8, 9). At least two-thirds of patients live independently (8±11) and low to very low rates of employment are reported regardless of source of recruitment (12, 13). Reported duration of illness (16 years) is similar in samples recruited from GPs and mental health services (5, 9, 11) and at least half display active signs of illness (positive and/or negative symptoms) in both groups (8±10, 14). At least half (15) and up to 98% (5) of patients are in contact with GPs regardless of recruitment source, but high rates of GP consultation do not imply less contact with public mental health services (16). The main unmet need in patients recruited from both sources is vocational rehabilitation (5, 13, 17). Thus, studies comparing schizophrenia patients in general practice with those in public mental health services are relatively few and no clear conclusions can be drawn. At present, there are no comparative data on schizophrenia patients recruited from general practice and public mental health services in terms of level of education, number of ospring, premorbid adjustment, mode of illness onset, level of disability, self-harm behaviours, patterns of substance abuse, prescribed medications or criminal history. The aim of the present study was to compare a sample of schizophrenia patients recruited from general practice with those recruited from public in-patient and community mental health services in terms of socio-demographic and clinical variables, levels of disability and patterns of service utilization. The main hypothesis was that the pro®le of the general practice sample would re¯ect an older, better functioning group with less severe illness, lower comorbidity and less service utilization. Material and methods Data sources
The data reported in this paper were drawn from two sources. The ®rst was a four-center Australian
survey of psychotic disorders conducted in 1997, known locally as the Low Prevalence (psychotic) Disorders Study (LPDS). The design, methods and an overview of the study's main ®ndings have been reported elsewhere (18). In that study, 980 individuals in contact with health services or marginal accommodation and aged 18±64 years were interviewed using the Diagnostic Interview for Psychosis (DIP), which records sociodemographic data, social functioning, symptom and diagnostic information, and service utilization (18). Of this sample, 456 participants recruited from public in-patient psychiatric facilities (n 178), public out-patient or community mental health services (n 236) and GPs (n 42) met ICD-10 diagnostic criteria for schizophrenia or schizoaective disorder. Participants who had been hospitalized for 6 months or more in the previous year were excluded. The second source of recruitment was through GPs in Newcastle and the surrounding Hunter region of New South Wales. This was part of a larger study of schizophrenia in general practice conducted in 1998±2000, which included an evaluation of GP needs in treating schizophrenia. All GPs in this region (n 493) were approached to assist with recruitment of schizophrenia patients in their care who were over the age of 17 years. There was an 88.4% (n 436) response rate from the GPs; 26.6% (n 131) reported that they were not treating any patient with schizophrenia; 25.2% (n 124) had such patients but refused to help recruit for the study; and 3.9% (n 19) replied that the schizophrenia patients they were treating were unsuitable for interview. This left 162 GPs (32.9%) who agreed to help recruit patients for the study. These GPs were given a total of approximately 510 `invitation letters' for distribution to the patients that they identi®ed as suitable for the study. Of the 155 patients who subsequently contacted us, 95 (61.3%) consented to an interview, 55 (35.5%) refused and 5 (3.7%) were deemed ineligible. Of those who were interviewed, 81 participants (85.3%) met ICD-10 diagnostic criteria for schizophrenia or schizoaective disorder on the basis of the DIP. Thus, for the present analyses, data were available for 123 GP-recruited subjects with schizophrenia or schizoaective disorder (42 from the LPDS and 81 from the associated Hunter region project); thereby providing a sucient sample size for comparisons with participants in the national study recruited from community mental health services (n 236) and psychiatric in-patient facilities (n 178). Overall, 86.0% of the selected subjects had an ICD-10 diagnosis of schizophrenia and 14.0% had schizoaective disorder; however, 347
Carr et al. there was a smaller percentage with schizoaective disorder (4.1%) in the GP-recruited sample. Prior to aggregating data for the two GP-recruited subgroups, preliminary analyses were conducted to con®rm that there were no signi®cant subgroup dierences with respect to age, gender, education, marital status, employment or service utilization. Measures
All subjects were administered the DIP (18), a semistructured interview comprising of three modules: (i) demographic and social functioning, including selected items from the WHO Disability Assessment Schedule (DAS) (19); (ii) diagnosis using the Operational Criteria for Psychosis (OPCRIT) (20) and elements of the WHO Schedules for Clinical Assessment in Neuropsychiatry (SCAN) (21); and (iii) usage of a range of hospitaland community-based health services in the past year. Apart from its role in con®rming clinical diagnoses, the DIP provided valuable data about socio-demographic characteristics, premorbid adjustment, substance use, current symptoms and medication, disability and social functioning, and service utilisation. While the DIP focused on lifetime substance abuse/dependence, an estimate of current substance use problems was also derived, comprising the percentage of participants with a lifetime history of substance abuse/dependence and hazardous consumption during the past year (i.e. daily use of alcohol or at least weekly use of any illicit substance). The inter-rater reliability of the DIP in the multicenter national study has been reported (generalized kappa 0.73 for the ICD-10 diagnoses) (18). In the Hunter project, interviewers were trained by personnel involved in the national study and achieved an inter-rater reliability (kappa) of 0.90 for the ICD-10 diagnoses, based on 29 interviews; for global ratings on the Social and Occupational Functioning Assessment Scale (SOFAS) (22), the corresponding inter-rater agreement (correlation) was 0.85. All of the items in the DIP covering current symptoms and mental state, and symptoms during the preceding 12 months, were subjected to a principal components analysis to help con®rm their patterns of association. Based on the item loadings, scores on four symptom factors were derived: depression (range: 0±20); mania (range: 0±9); reality distortion (range: 0±16); and disorganization (range: 0±11). To facilitate comparisons with other studies, a negative symptom subscale score (range: 0±3) was also derived, based on three of the disorganization items. 348
Similarly, two disability scales were constructed based on item loadings from a principal components analysis of the DAS: a personal disability score (range: 0±10), which covered ®ve DAS items (participation in household activities, interests, self-care, occupational performance and overall socializing); and a social disability score (range: 0±6), which included three DAS items (intimate relationships, deterioration in relationships and social withdrawal). We also report re-grouped global ratings (range: 0±10) from the SOFAS, with higher scores indicating better functioning. Data analysis
Group comparisons were undertaken using oneway analysis of variance (ANOVA) for the continuous dependent variables and v2 analysis for the categorical variables. As a partial control for the number of statistical tests, the threshold for signi®cance was set at P < 0.01. Results
Participants recruited from GP, community and in-patient sources were ®rst compared on a number of demographic and social variables, as shown in Table 1. General practice patients tended to be older than the in-patients and there were no dierences in gender distribution across the three groups. In-patients had a slightly younger age at illness onset than the community mental health sample. The general practice and community samples were both more highly educated than the in-patient sample, and a higher proportion of the general practice group were living in their own accommodation compared with the community group, who were more likely to be living in some form of supported accommodation. The general practice and community mental health groups were more likely than the in-patients to be employed. Those recruited from general practice also had higher levels of family support than the other two groups. There were no signi®cant group dierences in the percentage receiving a pension or other bene®t, the percentage with children and the mean availability of friends. Table 2 shows comparisons between the recruitment sources in terms of clinical and related variables. The general practice sample had better premorbid work adjustment than the community and in-patient samples. They also tended to have better premorbid social adjustment and a reduced likelihood of premorbid personality disorder, although these dierences were not statistically signi®cant. There were no group dierences in
Schizophrenia and recruitment source Table 1. Selected demographic and social comparisons between patients with schizophrenia or schizoaffective disorder recruited from general practice, community mental health services, and in-patient psychiatric units Source of recruitment Sociodemographic variable Sample size Age (SD) Male (%) Mean age at onset of mental illness (SD) Post-school qualifications (%) Own accommodation in past month (vs. supported, etc.) (%) Never married (%) Employed (full- or part-time) (%) Receiving benefits (%) With children (%) Mean face-to-face family support (0±6) (SD) Mean availability of friends (0±3) (SD)
General practice (G)
Community mental health services (C)
In-patient psychiatric units (I)
123 39.75 (11.56) 65.9 23.72 (8.05) 37.7 87.8
236 38.32 (11.48) 65.3 23.92 (7.40) 31.5 71.6
178 35.90 (11.37) 64.0 21.75 (7.44) 20.5 N/A
59.3 23.6 84.6 35.8 3.85 (1.59) 1.95 (0.89)
69.5 30.1 88.6 25.4 2.94 (1.75) 1.95 (0.87)
74.7 13.5 93.3 28.1 2.91 (1.74) 1.79 (1.03)
Pattern of significant differences between groups1 F(2, 534) = 4.45* v2(2) = 0.12, NS F(2, 534) = 4.62* v2(2) = 11.36* v2(1) = 12.03** v2(2) = 8.09, NS v2(2) = 15.77** v2(2) = 5.90, NS v2(2) = 4.28, NS F(2, 534) = 13.51** F(2, 525) = 1.85, NS
G>I C>I G, C > I G>C
G, C > I
G > C, I
1 Based on overall chi-square tests (categorical variables) or one-way analyses of variance (continuous variables), with pairwise (chi-square or ScheffØ) follow-up comparisons: NS, non-significant, *P < 0.01, **P < 0.001; N/A, not applicable, or value directly influenced by recruitment source.
mode of illness onset, lifetime substance abuse/ dependence comorbidity, or current substance use problems and smoking. In terms of symptoms, both the general practice and community groups had lower scores on reality distortion and disorganization, the general practice group had lower mania scores than the in-patients, and all three groups were signi®cantly dierent from each other in their negative symptom scores (general practice lowest, in-patients highest). The three groups were distinguishable from each other on all measures of disability, namely personal disability, social disability and SOFAS score. The general practice group was least disabled and the in-patient group most disabled. There were no group dierences in suicidal ideation, but the general practice and community samples were less likely to have attempted selfharm during the past year. The in-patient sample were more likely than the other two groups to have been arrested for an oence and there was a corresponding non-signi®cant trend (P 0.02) for a higher percentage of the in-patient group to have been charged with an oence. While the general practice and community groups were less likely to be dissatis®ed with their level of independence than the in-patients, there were no group dierences in overall dissatisfaction with life. The ®ndings with respect to service utilization and treatment are shown in Table 3. Participants recruited from general practice had fewer hospital admissions than those recruited from community services and both spent less time in hospital in the past year than the in-patient sample. Those recruited from general practice were also less likely to have contact with community mental
health services than those recruited from in-patient units. The general practice group had less contacts with community mental health services than the community mental health group. Users of rehabilitation services from the general practice and community samples spent more weeks in rehabilitation programs than those from the in-patient sample. Among those who contacted a GP during the last year, the mean number of GP visits was higher in the GP recruited sample than the other two groups. In terms of medication, the in-patient sample had a higher rate of use of clozapine, was more likely to be on more than one antipsychotic drug, and reported a higher rate of impairment caused by drug side-eects than the general practice and community samples. The general practice group was less likely to be taking a mood stabilising agent than the other two groups. There were no other group dierences in terms of medication. Both the general practice and community samples reported less unmet need for services than the in-patient sample. The general practice participants were subdivided according to whether they were treated by GPs alone (i.e. without the assistance of mental health services or individual psychiatrists or psychologists) or by GPs in association with specialist mental health services or practitioners. There were 48 participants (39.0%) in the former group and 75 in the latter (61.0%). Those being treated by GPs alone reported less family support (means: 3.25 vs. 4.23; F(1, 121) 11.97, P < 0.001), less depression (means: 1.63 vs. 4.56; F(1, 121) 13.60, P < 0.001) and less suicidal ideation (12.5 vs. 37.3%; v2(1) 9.02, P < 0.01). There were no other 349
Carr et al. Table 2. Selected comparisons of clinical and related variables between patients with schizophrenia or schizoaffective disorder recruited from general practice, community mental health services and in-patient psychiatric units Source of recruitment Clinical variable Sample size Premorbid adjustment (%) Poor premorbid social adjustment Poor premorbid work adjustment Premorbid personality disorder Mode of illness onset Mean mode of onset, abrupt vs. insidious (1±5) (SD) Lifetime substance abuse/dependence (%) History of alcohol abuse/dependence History of cannabis abuse/dependence History of other substance abuse/dependence Any history of substance abuse/dependence Currently smoking Current substance use problems Symptomatology (current or past year) Mean depression score (0±20) (SD) Mean mania score (0±9) (SD) Mean reality distortion score (0±16) (SD) Mean disorganisation score (0±11) (SD) Mean negative symptom subscale of disorganisation score (0±3) (SD)
General practice (G) 123
236
23.6 17.9 7.3
37.4 26.0 14.6 47.2 66.7 31.7
3.89 (1.36)
4.26 0.82 5.09 0.79 0.44
Pattern of significant differences between groups1
v2(2) = 6.92, NS v2(2) = 15.58** v2(2) = 7.81, NS
33.1 38.8 17.4
32.6 29.2 14.0 47.0 66.5 31.5 (4.52) (1.48) (4.03) (0.86) (0.52)
In-patient psychiatric units (I) 178
37.3 34.3 17.8
3.72 (1.52)
3.41 0.39 4.12 0.43 0.21
Community mental health services (C)
3.74 (1.37)
4.42 1.32 7.08 1.39 0.71
534)
v2(2) v2(2) v2(2) v2(2) v2(2) v2(2)
25.3 33.7 12.9 45.5 79.0 25.8 (5.05) (2.13) (3.98) (1.14) (0.70)
F(2,
(5.03) (2.66) (3.60) (1.75) (0.88)
F(2, F(2, F(2, F(2, F(2,
3.62 (1.95) 2.19 (1.43) 4.90 (1.79)
F(2, F(2, F(2,
= = = = = =
534) 534) 534) 534) 534)
= 0.78, NS
5.32, 2.16, 0.19, 0.12, 8.75, 1.87, = = = = =
NS NS NS NS NS NS
1.69, NS 6.69** 23.98** 20.51** 17.36**
Disability Mean personal disability (0±10) (SD) Mean social disability (0±6) (SD) Mean SOFAS score (0±10) (SD)
1.72 (1.76) 1.12 (1.35) 6.56 (1.30)
Self-harm in past year (%) Suicidal ideation Self-harm attempts
27.6 8.1
36.4 12.4
36.0 20.3
v2(2) = 3.12, NS v2(2) = 9.74*
3.3 3.3 8.9
8.1 5.5 11.6
14.0 11.2 23.2
v2(2) = 10.82* v2(2) = 8.48, NS v2(2) = 14.42**
15.7
13.8
22.8
v2(2) = 5.68, NS
Other (%) Arrested for an offence Charged with an offence Mostly dissatisfied with level of own independence Mostly dissatisfied with life
2.67 (1.71) 1.57 (1.42) 5.69 (1.41)
498) 507) 534)
G < C, I
= 38.16** = 20.52** = 43.10**
G
